Full Stack Engineer - AI

Full Stack EngineerDelamain AIJune 2026 · Hiring Requirement DocumentDelamain AI is an enterprise-grade AI intelligence platform serving federal and stategovernment affairs and policy advisory firms. This role owns the full stack from real-timedata pipelines and LLM summarization engines to the user-facing web portal.LocationIn-person, San FranciscoExperience3 – 5 yearsCompensation$150K – $180K + equityReports ToCEO & CTOAbout the PlatformDelamain AI is a custom AI platform supporting Washington, DC government affairs firms coveringtrade, defense, and financial services policy. The platform delivers automated intelligence and operationaltools to a team of policy advisors and principals.Morning Intelligence Briefings - Automated daily summaries generated from dozens of live policy sources— Truth Social, X, CNBC, Reuters, CRS Reports, Congressional Record, Punchbowl, foreign state media, and more.Secure Messaging - Signal CLI , WhatsApp or other chat integrations for end-to-end encryptedteam communications.AI Scouting Personas - Autonomous agents monitoring the policy landscape across trade, defense,and financial services domains.Research & Memo Agents - LLM-powered drafting and research assistance for policy memos and clientdeliverables.Scheduling Agent - Calendar and meeting management automation for principal staff.Unified Web Portal - UI interface for briefings, agents, and team management.Email: jimmy@daemo.aiBook Meeting - https://calendly.com/d/cyqq-jk9-csv/daemo-ai-hiring-callProgramming LanguagesTypeScript Expert — Required - Primary language across frontend (Next.js) . Allproduction code is TypeScript.Python Proficient — Required - Data pipelines, web scraping, CLI tooling, automationscripts, and LLM orchestration helpers.SQL / BigQuery SQL Proficient — Required - BigQuery-based data ingestion and querying;MongoDB aggregation pipelines for application data.Bash / Shell Working Knowledge - EC2 operations, deployment scripts, cron jobmanagement, and Linux service administration.Frameworks & LibrariesFrontendNext.js (App Router) Required — Expert Primary web framework. Proficient withserver components, server actions, API routes, and streaming.React Required — Expert Component architecture, hooks, server/clientcomponent boundaries.Tailwind CSS / shadcn/ui Required - Primary UI component and styling system usedacross all Delamain products.NextAuth.js Required - Authentication via Google OAuth; sessionmanagement, protected routes.Backend / APINode.js + Next.js Server Actions Required - REST and streaming endpoint design; OpenAI-compatible API clients for Delamain Engine integration.OpenAI-Compatible API (chat completions, tool use, streaming) Required — Expert Delamain Engine is an OpenAI-compatible LLM proxy. Must be fluent with tool/function calling, context management, and streaming responses.Signal CLI Preferred - Secure messaging integration deployed on Linux EC2; reliability and uptimemaintenance.Playwright / Puppeteer / BeautifulSoup Required - Web scraping from Congress.gov, Punchbowl, political news sources. Handling rate limits and anti-bot measures.RUST Proficient — Required - Rust powers the Delamain Engine — the central AI backend that everything routes through.Data & StorageMongoDB Atlas Required - Primary application database. Document modeling, aggregation pipelines, Atlas Search.Google BigQuery Preferred - Large-scale data ingestion, parallel table updates, policy data querying.Redis Preferred Caching, job queuing, pub/sub for pipeline coordination.InfrastructureDocker / Docker Compose Required - All services run in containers. Must be comfortable writing Dockerfiles and managing multi-service Compose configurations.AWS (EC2, Bedrock, S3) Preferred - Primary cloud. EC2 for hosted services; Bedrock forClaude LLM inference; S3 for file storage.Cloudflare Preferred - DNS, tunnels, access policies for production deployments.Linux / tmux Required Services run as persistent tmux sessions on EC2.Must be comfortable with SSH, process management, logs, and cron.CI/CD (GitHub Actions) Preferred - Automated build, test, and deploy pipelines.Engineering SkillsMust-Have• Full stack ownership — builds a feature end-to-end: data model ® API ® UI ® deploy. Does not hand off between tiers.• Async & concurrent systems — parallel data ingestion pipelines, concurrent BigQuery writes, racecondition awareness and prevention.• REST API design — clean, versioned, and documented APIs with consistent error handling.• Production debugging — reads logs across multiple pipeline stages, traces failures, and resolves issuesindependently.• Linux operations — SSHes into EC2, manages processes, writes cron jobs, diagnoses service failureswithout guidance.• Performance mindset — optimizes scraping throughput, pipeline latency, and LLM call cost/qualitytradeoffs.• CI/CD discipline — automated testing, build verification, and deployment pipelines.• Data pipeline architecture — ETL/ELT patterns, scheduling, failure handling, and deduplication acrossheterogeneous sources.• Multi-tenant / enterprise security — data isolation per client, access control, and audit logging patterns.• Web scraping at scale — anti-bot evasion, rate limiting, structured extraction from government and politicalsources.• Real-time streaming — WebSockets or SSE for streaming agent responses to the frontend.AI / ML SkillsMust-Have• LLM API integration — fluent with OpenAI-compatible APIs: chat completions, tool/function calling,streaming, context window management.• Prompt engineering — writes effective system prompts for intelligence summarization, persona-basedagents, and research tasks.• Agent architecture — understands multi-step tool use, memory patterns, and retrieval-augmentedgeneration (RAG).• AI output evaluation — identifies hallucinations, grades output quality, and implements guardrails inproduction pipelines.Strong Preference• AWS Bedrock — Claude 3.x model families (Haiku, Sonnet, Opus); model selection tradeoffs for cost vs.quality vs. latency.• Multi-provider LLM routing — Fireworks AI or similar; experience with fallback routing between inferenceproviders.• Embedding + vector search — Voyage AI or equivalent for semantic search over policy documents andbriefing history.• Automated briefing pipelines — scheduled agents that ingest from live sources, summarize, and deliverstructured output.Nice-to-Have• Fine-tuning or RLHF exposure• Named entity extraction and political entity resolution• Experience with LangChain, LlamaIndex, or custom agentic orchestration frameworksStrong PreferenceCandidates must demonstrate excellent skills and drive. It is a high-ownership position on a small,high-output engineering team.What This Means for You• You don't need a ticket to notice something is wrong — you find it, fix it, and document it.• You can hold a technical architecture conversation with the CTO and push back when warranted.• You operate with startup urgency. A 6-hour turnaround on a critical briefing pipeline failure is a baseline expectation.• You have strong opinions about code quality but know when to ship and refactor later.• You bring taste — the product quality bar is high, and it shows in everything from API design to UI polish.• You are not intimidated by ambiguity. You define it and build toward clarity.What You'll Work OnData Ingestion Pipeline - Scraping improvements and expanded coverage (Congress.gov, Punchbowl staffers, political media). Deduplication, BigQuery writes, scheduling.Morning Briefing Engine - LLM summarization pipeline — Bedrock (Claude) primary, Fireworks AIfallback. Delivery to portal and Signal.UI Web Portal - Next.js frontend features: user management, briefing display, agent interfaces, real-time streaming.Signal Integration - Reliability and uptime of EC2-hosted signal-cli for secure team messaging.Delamain Engine - Core architecture and infrastructure of the Delamain AI harness.This Role Is NOT Right For You If…• You need a detailed spec to start coding.• You have never debugged a production pipeline failure outside business hours.• You think 'full stack' means frontend or backend depending on the sprint.• You have never integrated an LLM API beyond a tutorial or demo project.• You want a large team around you before making architectural decisionsInterview ProcessWho / WhatInitial interview for Mutual fit followed most likely by a hackathon project and/or one week trial.